Comparing across regions
BaseScore is normalized so that locations in different cities and countries can be compared on a consistent scale. That is the design intent, and for most purposes it holds.
There is a caveat worth understanding, because teams that do not know about it tend to invent their own corrections.
What normalization does
BaseScore accounts for population density and area type, so a score is not simply a function of how busy a place is. A dense city center and a quiet suburb are each scored relative to comparable environments rather than against a single global count.
This is what makes a score of 72 in one city meaningful next to a score of 72 in another.
Where to apply judgment
Normalization operates on the underlying source data, and source data quality is not uniform across the world.
A city with comprehensive, well-classified law enforcement reporting produces a richer picture than a city where fewer sources publish and classification is inconsistent. BaseEngine models across those gaps, but modeled coverage and dense reported coverage are not identical inputs.
The practical consequence: cross-region comparisons are sound for ranking and tiering, and warrant more care when small differences are load-bearing. A 70 in one region against a 45 in another is a real difference. A 68 against a 71 across two very different data environments is not a distinction to build a decision on.
What to do about it
Rank and tier across regions — freely. Sorting a global portfolio by BaseScore to find your highest-risk sites is exactly what the score is for.
Check the data foundation before leaning on small differences. Open the Data Source module for both locations. If one is supported by many contributing sources and the other by few, weight the comparison accordingly.
Read the drivers, not just the number. Two locations at the same score with different category breakdowns represent different problems. The breakdown travels across regions better than the headline number does.
Say what you did. If you apply your own judgment on top of a score when briefing stakeholders, write that down. An adjustment nobody can see is an adjustment nobody can check.
If a score does not match what you know
If a location scores in a way that contradicts credible ground truth — your team has direct reporting that conflicts with the score — that is worth raising rather than working around.
Send it to your CSM with the location, the score, and what you are seeing on the ground. Ground-truth conflicts are among the most useful signals for improving the model, and a score that a team quietly distrusts is a score that stops getting used.